1Z0-184-25 Question 173
Single answerYou are tasked with loading a large CSV file containing text embeddings into an Oracle Autonomous Database table. The table has a column of type VECTOR(256) for storing 256-dimensional vector data. Each row in the CSV includes an identifier, a text description, and 256 floating-point values representing the embedding. Which approach ensures a successful load using SQL Loader?
- A
A. Specify the VECTOR(256) column in the control file as 256 separate comma-delimited columns of type FLOAT, then concatenate them into one column after loading.
- B
B. Define the VECTOR(256) column in the table definition and refer to it directly in the control file, using a single field with the correct delimiter and dimension.
- C
C. Convert the CSV vector data into a BASE64 string and load it as CLOB into the VECTOR(256) column, relying on implicit casting by SQL Loader.
- D
D. Load the entire row as a single JSON object and let SQL Loader automatically parse the VECTOR(256) column out of the embedded JSON structure.
Show answer and explanation
Correct answer: B
Explanation
In Oracle Database versions supporting VECTOR data types, you can directly define columns (e.g., VECTOR(256)) and then configure SQL Loader to handle these columns by specifying how the floating-point values in the CSV file map to the VECTOR column. Through the control file, you identify the field delimiter (commas in this scenario) and ensure that all the embedding values are treated as one logical field that matches the dimension of the VECTOR column. For more details, refer to the Oracle Database documentation on using SQL Loader and VECTOR data types in supported releases.
- A. Incorrect.
Option A: Incorrect. While it� possible to define 256 separate columns in some scenarios, you would not then concatenate them back into a VECTOR(256) column in a straightforward manner. This approach typically requires much more post-processing and defeats the purpose of leveraging the native VECTOR type directly.
- B. Correct.
Option B: Correct. By defining the VECTOR(256) column in the table and referencing it properly in the SQL Loader control file, you can load all floating-point values into a single VECTOR(256) column. SQL Loader supports mapping a single delimited field to the VECTOR type when configured correctly.
- C. Incorrect.
Option C: Incorrect. SQL Loader does not automatically convert a BASE64-encoded string into a VECTOR type. That would require custom parsing or application logic. Relying on implicit casting here is a misunderstanding of how SQL Loader handles data formats.
- D. Incorrect.
Option D: Incorrect. While JSON loading is supported for many scenarios, simply storing each row as a JSON object and expecting the VECTOR(256) to be parsed out automatically is not a standard SQL Loader feature. You would need additional steps or a different loading mechanism (like external tables with JSON support) to achieve that.